Smashbox in Python 3

The first half of the GSoC coding period is almost over, and this first entry to the developer journal has been long overdue. I’ll get right to the point:


I have fallen behind on the schedule I had proposed. Partly due to bad judgement, but mostly as my university academic schedule got super hectic because of COVID.


Over the last month, I have

  • Read through a large part of the codebase to understand the general lay of the land
  • Made a new YAML based configuration system: Configuration was earlier stored in a file, as valid Python code, which was then basically passed to exec1 to read all the settings in. Now, configuration is stored in a YAML file, that is read and parsed.
  • Migrated away from pycurl: Pycurl code is tough to read and maintain; we now use httpx (which comes with the added advantage of asyncio support)


Next, I’ll move on to

  • Updating the CLI code (maybe use something like the click library instead of stdlib argparse)
  • Update the multiprocessing engine 2
  • Migrate to loguru: loguru is an alternate logging library that is simpler to use than the stdlib logging, and gives prettier outputs. The codebase has a mix of logger.log and plain old print statements, so I’ll also standardize them (and only use the logger)


  • Getting smashbox itself running was quite difficult, I probably wouldn’t have been able to do it without the mentor’s explaining how to do it first. The existing README instructions have some out-of-date instructions. This could potentially be improved in my project; maybe also releasing a Docker image would make it easier to get started?
  • Working with a Python 2 codebase has been more difficult than I anticipated. I find myself constantly reaching for the documentation to find what some old deprecated function did so that I can find modern equivalents.

Hoping for an even more productive month of coding ahead! 🤞

  1. Well, execfile to be precise; a relic from Python 2 I hadn’t previously heard of 

  2. This is the part of the code that spins up multiple workers for test cases where we simulate multiple users interacting simultaneously.